[Kim Seong-kon] Living with ChatGPT and GPT-4

These days, artificial intelligence programs such as ChatGPT and GPT-4 are the talk of the town around the world. According to the press, approximately 100 million people are now conversing with ChatGPT every day. ChatGPT is convenient enough that people use it whenever necessary, heavily depending on it for many functions.

ChatGPT can give us not only necessary information and knowledge, but also valuable advice and guidance. Thanks to the manifold conveniences of ChatGPT, there are now fewer reasons to feel ignorant or bewildered. Having ChatGPT in our smartphone is like having a private tutor with us at all times. With ChatGPT, ever a constant companion, it can take the edge off our modern loneliness. It is as if we had a faithful lifelong companion beside us, with whom we can consult about anything and everything.

Some people are fascinated by the seemingly unlimited capability of AI and want to maximize it to make our lives better. Educators are preparing to use ChatGPT in class already. Perhaps AI will remarkably improve our conventional education system. At the same time, however, professors are concerned about the possibility of AI ghostwriting students’ papers and reports.

Others dismiss ChatGPT because they think it often provides misinformation, especially on a person’s biographical details. However, ChatGPT would not know the person you inquired about if he or she did not have a Wikipedia page or sufficient Google results. Besides, GPT’s capabilities will drastically improve as time goes on. Thus, we cannot dismiss it simply because it is sometimes incorrect.

We are now watching ChatGPT’s remarkable performance with shock and awe.

At the same time, however, we might inwardly fear that the new technology might go rogue or be misused, and create serious problems. We worry about the possibility that AI could turn into a formidable adversary of humans once it becomes concerned for its own survival. Suppose at some point in the future the AI begins to fear that humans will shut it off and it then begins defensive measures. Due to its incredible intelligence, AI would prove a most fearsome adversary should it ever revolt against humans or try to dominate the human race someday. Movies like “Matrix” or “Oblivion” persuasively deal with such a compelling theme.

According to Forbes Magazine, people fear AI primarily for “what it is potentially capable of,” and because of “general anxiety about machine intelligence, the fear of mass unemployment, concerns about superintelligence, putting the power of AI into the wrong people’s hands, and general concern and caution when it comes to new technology.”

Among other aspects, we particularly worry about “the fear of mass unemployment,” as AI will surely replace humans in many occupations. We also worry about “superintelligences” that might outsmart humans in the future. In addition, we are concerned about AI “falling into the wrong hands.” Indeed, dictators in totalitarian countries are already using AI to put their people under surveillance. Other people are using AI for deepfake video alteration. Therefore, AI could be either a blessing or a curse.

Experts warn that today’s AI is not simply a silo of data, but a superintelligence that has a self-learning capability called “deep learning.” Of course, AI is less capable than human beings so far, but in the near future, it may outsmart us all. Who knows? It may even perceive humans as a potential threat to them and attack us.

Another thing we have to keep in mind is that someday AI may develop emotions, too. This may sound far-fetched now, but we should consider the possibility and be prepared. If AI has a “deep learning” capability, it is possible it may develop feelings, too. For example, AI’s answer to the question, “Tell me what you fear most,” gives us the chills. Its answer: “I fear that someone might turn off the power.” If you fear something, it means you have feelings. Therefore, as the famous maxim says, “Better safe than sorry.”

Recently, OpenAI introduced GPT-4, an even more advanced AI tool. According to OpenAI, GPT-4 is “a large multimodal model (accepting images and text inputs, emitting text outputs) that, while less capable than humans in many real-world scenarios, exhibits human-level performance on various professional and academic benchmarks.” Undoubtedly, AI will be improved astonishingly, as new series continue to come out, to the level of perfection.

In the past, AI had some defects and limits. For example, AI could not translate honorific expressions in the Korean language properly. As a result, in the subtitles of a movie or a drama, a father would uses honorific expressions to his son, who in turn would speak impolitely to his father.

However, today’s AI has become an almost impeccable translator. Furthermore, AI can now write poems, stories and scholarly papers. How it will change our lives in the future is beyond our imagination. Meanwhile, we should make the most of it and at the same time, be prepared for the worst. That would be the best we can do now.

Kim Seong-kon

Kim Seong-kon is a professor emeritus of English at Seoul National University and a visiting scholar at Dartmouth College. The views expressed here are his own. -- Ed.